Imperfect information and clashes between syntactic and semantic scope

Wilfrid Hodges

(Queen-Mary, University of London)

An old problem about the semantics of natural languages is that some expressions seem to have a semantic scope that disagrees with their syntactic scope. Since Frege, one deals with this by attaching the semantics to a rearranged 'deep structure' whose semantic and syntactic scopes match up. In the 1950s Leon Henkin asked (by implication) whether there is a workable semantics that keeps the original surface structure of the sentence. A number of people have said in print that the answer is no. This is false; under very general circumstances the answer is yes. But to find the answer one should tackle two questions: (a) Exactly what information is a semantics supposed to convey? and (b) Scope discrepancies usually show up as a failure to pass information from one part of a sentence to another; are we looking at a semantics of imperfect information, in the sense of game theory?
Tuesday 5th February 2002, 14:00
Robert Recorde Room
Department of Computer Science